Lifecycle Hooks

Armature provides a comprehensive lifecycle hook system that allows you to execute code at specific points during the application lifecycle. This is inspired by NestJS and provides similar functionality for Rust web applications.

Table of Contents


Overview

Lifecycle hooks enable you to perform operations like:

  • Initialization: Connect to databases, start background tasks
  • Cleanup: Close connections, flush caches, stop workers
  • Monitoring: Log application state changes
  • Resource Management: Acquire and release resources safely

All lifecycle hooks are async and return a LifecycleResult:

pub type LifecycleResult = Result<(), Box<dyn std::error::Error + Send + Sync>>;

Available Hooks

OnModuleInit

Called after a module's dependencies are resolved but before the application is fully bootstrapped.

#[async_trait]
pub trait OnModuleInit: Send + Sync {
    async fn on_module_init(&self) -> LifecycleResult;
}

Use cases:

  • Initialize database connections
  • Load configuration
  • Set up caches
  • Start background tasks specific to the module

OnModuleDestroy

Called before a module is destroyed during application shutdown.

#[async_trait]
pub trait OnModuleDestroy: Send + Sync {
    async fn on_module_destroy(&self) -> LifecycleResult;
}

Use cases:

  • Close database connections
  • Flush caches
  • Stop background tasks
  • Clean up temporary resources

OnApplicationBootstrap

Called after all modules have been initialized and the application is fully ready.

#[async_trait]
pub trait OnApplicationBootstrap: Send + Sync {
    async fn on_application_bootstrap(&self) -> LifecycleResult;
}

Use cases:

  • Perform post-initialization setup
  • Start global services
  • Log application readiness
  • Trigger initial data synchronization

OnApplicationShutdown

Called during graceful application shutdown.

#[async_trait]
pub trait OnApplicationShutdown: Send + Sync {
    async fn on_application_shutdown(&self, signal: Option<String>) -> LifecycleResult;
}

Use cases:

  • Gracefully terminate long-running operations
  • Send final metrics/logs
  • Notify external systems of shutdown
  • Save application state

BeforeApplicationShutdown

Called before the main shutdown hooks, allowing for pre-shutdown operations.

#[async_trait]
pub trait BeforeApplicationShutdown: Send + Sync {
    async fn before_application_shutdown(&self, signal: Option<String>) -> LifecycleResult;
}

Use cases:

  • Stop accepting new requests
  • Drain request queues
  • Notify load balancers
  • Prepare for shutdown

Hook Execution Order

Startup Sequence

1. Module Registration
   โ””โ”€> Providers and controllers registered in DI container

2. OnModuleInit
   โ””โ”€> Called for each service/controller (FIFO order)

3. OnApplicationBootstrap
   โ””โ”€> Called after all modules initialized (FIFO order)

4. Application Ready
   โ””โ”€> Server starts accepting requests

Shutdown Sequence

1. Shutdown Signal Received
   โ””โ”€> SIGTERM, SIGINT, or manual shutdown

2. BeforeApplicationShutdown
   โ””โ”€> Called for pre-shutdown operations (FIFO order)

3. OnApplicationShutdown
   โ””โ”€> Called for graceful shutdown (LIFO/reverse order)

4. OnModuleDestroy
   โ””โ”€> Called for cleanup (LIFO/reverse order)

5. Application Terminated

Important: Destroy and shutdown hooks are called in reverse order (LIFO) to ensure proper cleanup of dependencies.


Usage

Basic Implementation

Implement lifecycle hooks on your services or controllers:

use armature_core::{Provider, lifecycle::{OnModuleInit, OnModuleDestroy}};
use async_trait::async_trait;

struct DatabaseService {
    connection: Option<Connection>,
}

impl Provider for DatabaseService {}

#[async_trait]
impl OnModuleInit for DatabaseService {
    async fn on_module_init(&self) -> LifecycleResult {
        println!("Connecting to database...");
        // Initialize database connection
        Ok(())
    }
}

#[async_trait]
impl OnModuleDestroy for DatabaseService {
    async fn on_module_destroy(&self) -> LifecycleResult {
        println!("Closing database connection...");
        // Close database connection
        Ok(())
    }
}

Registration with Lifecycle Manager

use armature_core::LifecycleManager;
use std::sync::Arc;

let lifecycle = LifecycleManager::new();
let db_service = Arc::new(DatabaseService { connection: None });

// Register hooks
lifecycle.register_on_init("DatabaseService".to_string(), db_service.clone()).await;
lifecycle.register_on_destroy("DatabaseService".to_string(), db_service).await;

Integration with Application

The lifecycle manager is integrated into Application:

use armature_core::Application;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create application (hooks are called automatically)
    let app = Application::create::<AppModule>().await;

    // Application runs...

    // Graceful shutdown
    app.shutdown(Some("SIGTERM".to_string())).await?;

    Ok(())
}

Error Handling

Lifecycle hooks can return errors, which are collected and reported:

#[async_trait]
impl OnModuleInit for MyService {
    async fn on_module_init(&self) -> LifecycleResult {
        if let Err(e) = self.connect().await {
            return Err(format!("Failed to connect: {}", e).into());
        }
        Ok(())
    }
}

Errors don't stop the application lifecycle - all hooks are called, and errors are logged:

๐Ÿ”„ Calling module initialization hooks...
  โœ— MyService: onModuleInit() failed: Failed to connect: Connection refused
  โœ“ OtherService: onModuleInit() completed

Best Practices

โœ… Do's

  1. Keep hooks fast: Lifecycle hooks should complete quickly
  2. Handle errors gracefully: Return meaningful errors
  3. Use appropriate hooks: Choose the right hook for your use case
  4. Clean up resources: Always implement both init and destroy if needed
  5. Log operations: Provide visibility into what's happening
  6. Make hooks idempotent: Hooks should be safe to call multiple times

โŒ Don'ts

  1. Don't perform long-running operations: Startup should be fast
  2. Don't ignore errors: Always handle and return errors properly
  3. Don't assume order: Don't rely on specific hook execution order
  4. Don't block: Use async operations, not blocking I/O
  5. Don't panic: Return errors instead of panicking

Idempotency Example

struct CacheService {
    initialized: Arc<RwLock<bool>>,
}

#[async_trait]
impl OnModuleInit for CacheService {
    async fn on_module_init(&self) -> LifecycleResult {
        let mut init = self.initialized.write().await;

        // Guard against multiple initializations
        if *init {
            println!("Cache already initialized, skipping");
            return Ok(());
        }

        // Initialize cache
        println!("Initializing cache...");
        *init = true;
        Ok(())
    }
}

Examples

Example 1: Database Connection Service

use armature_core::{Provider, lifecycle::{OnModuleInit, OnModuleDestroy}};
use async_trait::async_trait;
use std::sync::Arc;
use tokio::sync::RwLock;

struct DatabaseService {
    connection_string: String,
    pool: Arc<RwLock<Option<ConnectionPool>>>,
}

impl Provider for DatabaseService {}

#[async_trait]
impl OnModuleInit for DatabaseService {
    async fn on_module_init(&self) -> LifecycleResult {
        println!("๐Ÿ“Š Connecting to database: {}", self.connection_string);

        // Create connection pool
        let pool = create_pool(&self.connection_string).await?;
        *self.pool.write().await = Some(pool);

        println!("โœ… Database connection established");
        Ok(())
    }
}

#[async_trait]
impl OnModuleDestroy for DatabaseService {
    async fn on_module_destroy(&self) -> LifecycleResult {
        println!("๐Ÿ“Š Closing database connections...");

        // Close pool
        if let Some(pool) = self.pool.write().await.take() {
            pool.close().await?;
        }

        println!("โœ… Database connections closed");
        Ok(())
    }
}

Example 2: Background Worker

use armature_core::{Provider, lifecycle::{OnApplicationBootstrap, OnApplicationShutdown}};
use async_trait::async_trait;
use std::sync::Arc;
use tokio::sync::RwLock;

struct WorkerService {
    running: Arc<RwLock<bool>>,
    handle: Arc<RwLock<Option<tokio::task::JoinHandle<()>>>>,
}

impl Provider for WorkerService {}

#[async_trait]
impl OnApplicationBootstrap for WorkerService {
    async fn on_application_bootstrap(&self) -> LifecycleResult {
        println!("๐Ÿ”„ Starting background worker...");

        *self.running.write().await = true;
        let running = self.running.clone();

        let handle = tokio::spawn(async move {
            while *running.read().await {
                // Do work
                tokio::time::sleep(tokio::time::Duration::from_secs(1)).await;
            }
        });

        *self.handle.write().await = Some(handle);
        println!("โœ… Background worker started");
        Ok(())
    }
}

#[async_trait]
impl OnApplicationShutdown for WorkerService {
    async fn on_application_shutdown(&self, signal: Option<String>) -> LifecycleResult {
        if let Some(sig) = signal {
            println!("๐Ÿ›‘ Stopping worker (signal: {})...", sig);
        } else {
            println!("๐Ÿ›‘ Stopping worker...");
        }

        // Signal worker to stop
        *self.running.write().await = false;

        // Wait for worker to finish
        if let Some(handle) = self.handle.write().await.take() {
            handle.await?;
        }

        println!("โœ… Worker stopped gracefully");
        Ok(())
    }
}

Example 3: Health Check Service

use armature_core::{Provider, lifecycle::{OnApplicationBootstrap, BeforeApplicationShutdown}};
use async_trait::async_trait;

struct HealthCheckService {
    endpoint: String,
}

impl Provider for HealthCheckService {}

#[async_trait]
impl OnApplicationBootstrap for HealthCheckService {
    async fn on_application_bootstrap(&self) -> LifecycleResult {
        println!("โœ… Application ready - health checks enabled");

        // Notify load balancer that we're ready
        self.notify_ready().await?;

        Ok(())
    }
}

#[async_trait]
impl BeforeApplicationShutdown for HealthCheckService {
    async fn before_application_shutdown(&self, _signal: Option<String>) -> LifecycleResult {
        println!("โš ๏ธ  Marking application as unhealthy...");

        // Notify load balancer to stop sending traffic
        self.notify_shutting_down().await?;

        // Wait for existing connections to drain
        tokio::time::sleep(tokio::time::Duration::from_secs(5)).await;

        println!("โœ… Application marked unhealthy, connections drained");
        Ok(())
    }
}

Example 4: Multiple Hooks on One Service

A service can implement multiple lifecycle hooks:

struct ComprehensiveService {
    name: String,
    initialized: Arc<RwLock<bool>>,
}

impl Provider for ComprehensiveService {}

#[async_trait]
impl OnModuleInit for ComprehensiveService {
    async fn on_module_init(&self) -> LifecycleResult {
        println!("{}: Module initialization", self.name);
        *self.initialized.write().await = true;
        Ok(())
    }
}

#[async_trait]
impl OnApplicationBootstrap for ComprehensiveService {
    async fn on_application_bootstrap(&self) -> LifecycleResult {
        println!("{}: Application bootstrap complete", self.name);
        Ok(())
    }
}

#[async_trait]
impl BeforeApplicationShutdown for ComprehensiveService {
    async fn before_application_shutdown(&self, signal: Option<String>) -> LifecycleResult {
        println!("{}: Preparing for shutdown: {:?}", self.name, signal);
        Ok(())
    }
}

#[async_trait]
impl OnApplicationShutdown for ComprehensiveService {
    async fn on_application_shutdown(&self, _signal: Option<String>) -> LifecycleResult {
        println!("{}: Shutting down", self.name);
        Ok(())
    }
}

#[async_trait]
impl OnModuleDestroy for ComprehensiveService {
    async fn on_module_destroy(&self) -> LifecycleResult {
        println!("{}: Module cleanup", self.name);
        *self.initialized.write().await = false;
        Ok(())
    }
}

Signal Handling

The lifecycle system supports passing shutdown signals to hooks:

use tokio::signal;

async fn run_with_signal_handling(app: Application) -> Result<(), Box<dyn std::error::Error>> {
    // Wait for shutdown signal
    tokio::select! {
        _ = signal::ctrl_c() => {
            println!("Received Ctrl+C");
            app.shutdown(Some("SIGINT".to_string())).await?;
        }
        _ = wait_for_sigterm() => {
            println!("Received SIGTERM");
            app.shutdown(Some("SIGTERM".to_string())).await?;
        }
    }

    Ok(())
}

Testing Lifecycle Hooks

You can test lifecycle hooks directly:

#[tokio::test]
async fn test_service_lifecycle() {
    let service = Arc::new(MyService::new());

    // Test initialization
    assert!(service.on_module_init().await.is_ok());
    assert!(service.is_initialized().await);

    // Test cleanup
    assert!(service.on_module_destroy().await.is_ok());
    assert!(!service.is_initialized().await);
}

Advanced Usage

Conditional Hooks

You can implement conditional logic in hooks:

#[async_trait]
impl OnModuleInit for MyService {
    async fn on_module_init(&self) -> LifecycleResult {
        if std::env::var("SKIP_INIT").is_ok() {
            println!("Skipping initialization (SKIP_INIT set)");
            return Ok(());
        }

        // Normal initialization
        self.initialize().await?;
        Ok(())
    }
}

Timeout Protection

Add timeouts to prevent hooks from hanging:

use tokio::time::{timeout, Duration};

#[async_trait]
impl OnModuleInit for MyService {
    async fn on_module_init(&self) -> LifecycleResult {
        match timeout(Duration::from_secs(30), self.initialize()).await {
            Ok(Ok(())) => Ok(()),
            Ok(Err(e)) => Err(e),
            Err(_) => Err("Initialization timeout".into()),
        }
    }
}

Summary

Lifecycle hooks in Armature provide a powerful way to manage application state and resources:

  • โœ… 5 hook types for different lifecycle phases
  • โœ… Async by default for modern Rust applications
  • โœ… Error handling with Result types
  • โœ… Automatic execution by the Application
  • โœ… FIFO/LIFO ordering for proper initialization and cleanup
  • โœ… Signal support for graceful shutdown
  • โœ… Testable lifecycle logic

Use lifecycle hooks to build robust, maintainable Rust web applications with proper resource management! ๐Ÿš€